openforis / accuracy-assessment Goto Github PK

I have uploaded my map in the stratified area estimator design. Interestingly, the map code values displayed do not correspond to the map codes of my map classes. This is making it difficult for me to know which class is which.
issue.docx

While generating the point. The tool freezes and the message is displayed "reconnecting to server".

Hi,

Thanks for the hard work. Some suggestions and one issue:

One issue: I have uploaded a raster and I follow all steps, the work fine, but when I click on the generation of random points, it does not work. It shows: "An error has occurred. Check your logs or contact the app author for clarification."

One suggestion: We are taking about area estimation, so using the UA or the required standard error for the OA to estimate the number of sample are not very intuitive. I would suggest that instead of UA you call it "proportion of the variable of interest" and that se of OA, is the target standard error expressed in proportion of area.

Another suggestion: Different strata will have different proportions for a variable of interest given (maybe estimated via a pilot sampling or a previous excercise). It would be good to give the flexibility to the user to define the proportions for each stratum so that we have specific variances for each variable of interest.

Final one: Also give some guidance when we have multiple variables of interest. We might be specifying the best allocation for deforestation, but it might be a poor allocation for gains (e.x. we only allocate 100 samples in the gain stratum).

I am able to download the points as a spreadsheet, but not as a cep or shapefile using the stratified area estimator design tool. However the csv will download.

          Dear Daniel

Can we discuss quickly to fix this ?
Thanks

Originally posted by @lecrabe in #16 (comment)

Hi,

I was hoping to download the stratified random sample on the map but it gives me an error message.

Thanks,

Florence

When loading the module you get an error

Error in dyn.load(file, DLLpath = DLLpath, ...) :
unable to load shared object '/usr/local/lib/R/site-library/httpuv/libs/httpuv.so':
/usr/lib/x86_64-linux-gnu/libstdc++.so.6: version `GLIBCXX_3.4.29' not found (required by /usr/local/lib/R/site-library/httpuv/libs/httpuv.so)

Package HTTPUV is missing

The module requires updating, as the majority of R spatial packages were discontinued on November 23 and are no longer accessible on CRAN. Details are here.

We changed the structure of the images in the docs so all the images of the rst documentation of this module are broken

Hello,

I'm using the Stratified Area Estimator - Design for binary classification; at the final step of sample allocation, I got the invalid first argument. What might be the cause of the error, please.
I look forward to your response.
Thanks

Hi,

I just get an error "Failed-Server problem" when downloading as Collect Earth Online project.

Thanks,

Florence

Hi,

Nothing happens when clicking on "Create CEO project" on the last page of Sepal.

Thanks,

Florence

How can I upload my own map for the accuracy assessment design?

Hi,

I've been a user of the accuracy assessment tool of openForis for a while now. I would like to suggest some new options/features:

• it would be great if there would be an option to introduce more ways of stratifying, for instance, uploading a map where the classes are used as strata and and a polygon file with administrative borders that also can be used as strata. It would be even better if the amount of classes for stratification can be set per level of administration (f.i: 8 sub classes on the whole map and 2 global classes per sublevel of administration).

• another feature would be to have an option to select more 'pure' samples, in the sense that a pixel is only chosen as a sample for a certain class if all bordering pixels have the same class. I tried to simulate this with resampling the input map to a coarser resolution, but still too many edge cases exist.

thank you for all your effort, the tool has been put into good use.
all the best,

s.

hi,

I've used the aa_design shiny web application some months ago. It worked flawlessy and helped me calculating the number of and selecting samplings.

I tried to use it back today. I had a little trouble since the readme doesn't refer anymore to running the code with shiny straight from github. Luckily I found the needed R code in the history.
When the application opens in the browsers everything looks fine. But when selecting a 'Map Input' the popup browse screen remains blank. There is no selecting or browsing possible.

I tried with Firefox53 and IE11.

I've tried updating R (3.4) and RStudio (1.0143) to the latest version. I also tried after installing Rtools (34) which I can't remember being needed before.

Furthermore, If you are not supporting this kind of use anymore, would it be possible to get an account for SEPAL. Although I'm not in need of cloud processing (the data is not large).